BACKGROUND: Bupivacaine induces cardiac depression, which is resistant to treatment. Therefore early recognition of its development is important so that the injection of bupivacaine can be discontinued promptly. We compared the efficacy of continuous mixed venous oxygen saturation (cSvO2) monitoring with that of the mean arterial blood pressure (MBP) monitoring in terms of the prediction of cardiac output (CO) changes in anesthetized dogs with bupivacaine-induced cardiovascular depression. METHODS: Bupivacaine was infused to pentobarbital-anesthetized mongrel dogs (n = 8) at a rate of 0.5 mg/kg/min until the MBP decreased to 40 mmHg or less (end of bupivacaine infusion; BIE). We defined the early period as the interval from baseline till 30 minutes after the bupivacaine infusion and the late period as that after the 30 minutes till BIE. We monitored cSvO2 with a fiberoptic pulmonary artery catheter. The MBP and CO were measured every ten minutes after the initiation of the bupivacaine infusion. Arterial blood gas analysis, mixed venous gas analysis, measurement of serum electrolyte concentrations and serum bupivacaine concentrations were performed simultaneously. The relationship between CO versus cSvO2 or versus MBP was compared retrospectively by regression analysis. RESULTS: The Pearson's correlation coefficients between CO and cSvO2 were 0.782 (r2 = 0.6111, P < 0.01) in the early period and 0.824 (r2 = 0.6790, P < 0.01) in the late period. The correlation coefficients between CO and MBP were 0.019 (r2 = 0.0003, P > 0.05) in the early period and 0.799 (r2 = 0.6381, P < 0.01) in the late period. CONCLUSIONS: cSvO2 is superior to MBP for the prediction of CO changes in bupivacaine-induced cardiac depression in dogs, especially in the early period. We might expect a profound reduction of CO when MBP begins to decrease in the late period.